Yongji Fu

About

I am Yongji Fu (符永骥), an MSc student in Robotics Engineering at the University of Bristol (2025.09 – 2026.10), advised by Nathan F. Lepora and Guanqun Cao. Before Bristol I received my BSc in Information Management and Information Systems from Chongqing University of Posts and Telecommunications.

Goal

To build robotic and agentic systems that continuously learn and iteratively self-improve through interaction with the physical world.

Research Interests

large-scale machine learning · world model for robot learning · continuous self-evolving agent · general-purpose loco-manipulation

Experience

  • Research Intern, Bristol Robotics Laboratory, University of Bristol, 2025.09 – present. Continual-learning interactive robot, and visuo-tactile latent world models, under Nathan F. Lepora and Guanqun Cao.

  • Research Intern, Institute of Engineering and Applied Technology, Fudan University, 2025.07 – present. Learning realistic expressions for humanoid face robots — retargeting from human reference into the robot’s actuator space and a controller that balances visual fidelity with the hardware’s mechanical limits.

  • Algorithm Researcher, Chongqing Robotics Institute, 2023.08 – 2025.06. Led three industry-facing ML projects:

    • Legal-domain LLM assistant (with Southwest University of Political Science and Law and Beijing Chaoxing Tianxia): RoPE-extended long-context backbone, legal knowledge graph + RAG, tool-calling agent, and a small intent/NER model rewriting questions into formal logical symbols for reasoning.
    • Packaging QA for a hazardous-explosive production line (with Shaanxi North Civil Explosives Group): robust detection–segmentation dual-task network, buffered Cython rule gate, structural reparameterisation, and TensorRT deployment — ≥ 99% accuracy over 30 days in production, 4 → 15 FPS on an RTX 4060.
    • GNN-accelerated MILP scheduling for industrial electroplating: GNN warm-start + FENNEL partitioning + high-confidence variable fixing, yielding > 10× average speed-up.

  • Research Assistant / Team Lead, Big Data Intelligent Computing Lab, CQUPT, 2023.09 – 2024.12. Led the tennis-scoring and university-research-QA projects; filed 2 invention patents. Ran an annual AI/ML training programme for 40+ students.

Selected Awards

  • 5th place globally (of 432 teams), ByteDance AI Safety Challenge.
  • National Third Prize, 7th China Collegiate Computing Contest — Network Technology Challenge.
  • National Second Prize, 11th CAAI Digital Media Competition.

Skills

  • Languages & tooling: Python, C++, shell; Git, Weights & Biases; Markdown, LaTeX.
  • Deep learning: PyTorch, TensorFlow, JAX.
  • Deployment: CUDA, TensorRT, Triton — full training-to-deployment pipeline.
  • Systems: Linux, shell scripting.
  • English: TOEFL 103; comfortable reading English technical documentation and reproducing state-of-the-art research papers.

Contact

Email: yongji.fu7@gmail.com
GitHub: @yongjifu7